By The Copilot & Musso Media Editor
What do 1,000 journalists and PR pros know about AI that you don't? They took AI Quick Start, a 1-hour live class from The Media Copilot. 94% satisfaction. Find out how to work smarter with AI in just 60 minutes. Get 20% off with the code AIPRO: https://mediacopilot.ai/
Can AI help rebuild local journalism before the economics of the industry completely break? Axios does.
On this episode of The Media Copilot podcast, Pete Pachal speaks to Allison Murphy, COO of Axios, to explore how Axios Local is experimenting with AI-driven newsroom operations, “hyperlocal” expansion, and lean reporting teams designed to scale across hundreds of communities.
The needle to thread: how to produce high-quality local journalism without sacrificing editorial standards. From AI-assisted social publishing and newsroom training to experimental tools like the “Axiomizer” and “Localizer,” the conversation goes beyond vague AI hype and into the real mechanics of how modern media organizations are adapting in real time.
Murphy also explores newsroom trust, AI transparency, audience skepticism, regional expansion strategy, and the growing financial pressure forcing publishers to rethink how journalism can remain sustainable in the modern media economy.
“The fundamental challenge with local journalism now is a financial one. We are looking at how we can bring the cost of delivering really high quality, originally reported journalism and news and information to many, many communities.”
Murphy argues that the future of local news depends on finding the right balance between human expertise and technological efficiency before the economics of the industry become impossible to sustain.
What we cover:
• Why Axios sees AI as essential to saving local journalism economics
• How one reporter and “half reporter” cities are changing newsroom models
• The role of AI in reporting workflows, editing, planning, and social distribution
• Why Axios believes human reporters remain core to its journalism
• The rise of AI-enabled newsroom operations and internal employee training
• What media companies still misunderstand about AI adoption
• Reader trust, transparency, and AI disclosure experiments
• The future of audience growth and media discovery in an AI ecosystem
• Why local journalism may become more scalable than ever before
Listen or watch:
Why this matters:
Most conversations about AI in the media stay theoretical. This one gets operational.
Axios is actively testing what an AI-enabled newsroom looks like at scale—not in a lab, but across real communities with real reporters and real business pressures. As local journalism continues to shrink nationwide, the stakes are bigger than newsroom efficiency. They’re about whether sustainable local reporting can exist at all in the next decade.
For anyone working in journalism, media strategy, publishing, audience growth, or AI product development, this episode offers one of the clearest looks yet at how a modern newsroom is trying to adapt before the industry’s financial realities force even harder choices.
About the 👤 Guest
• LinkedIn: Allison Murphy
• Axios official site: Axios
About the show: To explore more conversations like this and see what’s new, visit the Media Copilot website at mediacopilot.ai. You’ll find new episodes, expanded resources, and tools designed for journalists, communicators, and media leaders navigating the fast-changing world of AI. It’s the home base for everything Media Copilot and it’s just getting started.
Enjoyed this episode?
Subscribe to The Media Copilot on Substack, Apple Podcasts, Spotify, or your favorite app. On YouTube? Tap the Like button and Subscribe to the YouTube channel. For more AI tools and resources built for media professionals, visit mediacopilot.ai.Produced by Pete Pachal and Executive Producer Michele Musso
Edited by the Musso Media Team Music: “Favorite” by Alexander Nakarada, licensed under CC BY 4.0
All rights reserved. © AnyWho Media 2026.
TRANSCRIPT
Pete Pachal ( 00:28.000)
Hi, welcome to the Media Copilot. It’s a podcast about how AI is changing media, news, and communication. I’m your host, Pete Pachal. I covered tech for a long time as a journalist. And now I talk with the media leaders, the builders, and the creators trying to answer the question, how will we get information in the future? And how will that transform journalism and the business of media?
Quick housekeeping note, if you’re listening on Apple or Spotify, please leave a five star review and maybe a nice comment. And if you’re watching on YouTube, please like the video and subscribe to the channel if you don’t mind. Those things really do help more people find the show.
My guest today is Alison Murphy, Chief Operating Officer of Axios. She runs Axios Local, which is one of the most interesting experiments in local journalism right now. It sits right at the intersection of three very important questions in media, whether local news can scale, whether small teams can produce high quality journalism that people want, and whether AI can help make that business work without weakening the product.
Axios Loco is already in dozens of cities and it’s expanding to many more this year. And more importantly, it’s using AI in ways that are much more specific than the usual vague talk about efficiency or productivity.They’ve even given one of their tools a name, the Axiomizer.
So today, I want to talk with Allison about how AI actually changes a local news operation where Axios draws the line on AI use and how we should judge whether or not this is all working. Should be a breeze. Allison Murphy, welcome to the Media Copilot.
Allison Murphy (02:10.414)
Thanks so much for having me, Pete. And I would like to say I did not come up with the name Axiomizer, so don’t put that one on me. It was named by our journalists, and I know that it’s a mouthful.
Pete Pachal (02:22.989)
So I think I might’ve mispronounced it. I left out one of the I’s. Axiomizer, not the ax-omizer. Yeah, I’ve gotta inventory them all at some point. So listen, before we get into all of it, I wanna dive into everything I just described, but I’d love to hear just a little bit more about you, your background, your role at Axios, so we can better understand where you’re coming from. So who are you?
Allison Murphy (02:26.471)
I mean, that’s, yeah, there’s a lot of vowels in there.
Allison Murphy (02:44.59)
Yeah, so I’ve spent about 15 years in media now. Began my media career at the New York Times where I did a number of different roles all focused on digital transformation. This is in the 2010s when so much effort was going towards building a sustainable subscription and advertising business built on digital at the Times. So I had a fantastic time being part of creating new technology-based solutions.
And thinking about the company as a whole and then also working for a while on data and ad products that could be privacy first and that really leaned into the relationship the Times had with readers. I was eager for new experiences and had always loved Axios as a reader and so when Axios sold as a company and started thinking about its next phase of growth. They were looking for some seasoned operators and I came in as the first kind of executive level operations leader and am now COO. And it’s a fantastic job because I get to do a little bit of everything. I oversee some of our GNA functions. I really work at the center of the company to make sure everything we’re doing is aligned between departments and newsroom because at any media company, there is no excess anymore. You need every effort, every person, everything you do to be focused and to be oriented towards the things that will drive success.
Pete Pachal (04:21.285)
Ha
Allison Murphy (04:32.864)
Most recently, of course, I oversee local and am now also because of local and because of where Axios sees the future going. I’m also deeply engaged in what we’re doing with AI enablement as a company with our employees and how we work, but also the technology we’re building that we think is going to enhance our journalism and how we serve readers and how we build a great media business.
Pete Pachal (04:59.329)
Yeah, I like how you sort of phrase that. Honestly, while you talked about not leaving anything on the table, it’s almost like AI in a nutshell, right? It’s like forcing this efficiency, especially in the media where, you know, it’s always been kind of a low margin business on, you know, not, it’s a broad brush I’m painting there, I realize, but it is, yeah. So yeah, I definitely want to talk about the local stuff. Like it seems that Axios Local, it’s a really interesting phase right now. Can you zero in on that a bit? Like how do you define…Like what is the business trying to do? What’s it sort of trying to prove over the next year or two as well? Cause it’s not just like business goals. It’s like this deep unknown of AI and like, is this going to work? So let’s start at the top level. Like what are you guys trying to prove over the next couple of years with respect to that?
Allison Murphy (05:43.062)
Yes.Absolutely.
Allison Murphy (05:46.935)
I think it’s no surprise to anyone that local journalism is in decline and decline is a gentle way of putting in it. Full on freefall would be a more accurate description. And what we are trying to do is show that there is at least a model to bring profitable and sustainable business to hundreds of communities, thousands of communities around America. And by the way, if you hear my dog in the background….
Pete Pachal (06:16.3)
Yeah, that’s right. Allison is not joining us from a kennel, ladies and gentlemen. An excited dog.
Allison Murphy (06:16.848)
You can tell that she’s enthusiastic about this topic as well. My, apologies. There’s a lot of construction in my neighborhood and that connects with local as well. meaning, what we aim to do is to be an essential daily guide for smart professionals in their communities. That means covering things like zoning and construction and schools, but also local politics, also restaurant openings.
We are not trying to recreate the full and important newsrooms of local newspapers of the past. What we see is that the fundamental challenge with local journalism now is a financial one. So we are looking at how can we bring the cost of delivering really high quality, originally reported journalism and news and information to many, many communities. We have a product that we know readers love.
Pete Pachal (07:19.223)
Mm-hmm.
Allison Murphy (07:22.48)
It’s a daily newsletter. It’s also on web and on social. It’s a lean model. There’s two or one, or we’re actually now beginning with kind of part-time partial reporters in some smaller geographies. So there is an essential human element of a reporter who understands that area.
But we see technology as being imperative to how we can keep the cost structure sustainable, how we can make sure that reporters can do this job because it’s not easy to be responsible for an entire product all on your own every day. And we think there’s a monetization angle to it. So the technology piece for us is is inextricable from what we’re trying to do, which is scale something high quality.
Pete Pachal (08:12.065)
Right, what is, now that AI is sort of in the picture, and again, I wanna get into the sort of details as well, to the extent that you can get into the weeds of it, because I love doing that, but I also wanna understand just at the outset, if you decide to go into a community, there must be sort of like kind of an absolute minimum that you would need to just kind of even be there. And you mentioned part-time, I mean, is there almost like, because some of these other companies that are trying to sort of crack the….
Allison Murphy (08:20.461)
Yeah.
Allison Murphy (08:23.937)
Yeah.
Pete Pachal (08:40.813)
code on using AI to enhance local coverage might even operating almost like headless way. I don’t think they use that term, but that’s kind of how I think of it when one of these sort of newsletter companies is like, yeah, we’re going to aggregate and we’re going to assemble certain things with AI and serve this community. They have different models. They’re not necessarily journalism first, or if they are, it might be a different way. And we can get into that. But I’m curious, what is Axios when they decide, OK, we want to do something in this community? What’s the absolute minimum now, and how has that changed?
Versus like, you know, three or four years ago when AI wasn’t in the picture.
Allison Murphy (09:12.822)
Yeah, it’s a great question. And some of that is about AI and some of it is about the nature of communities. So you’re correct that there are some interesting companies who are kind of seeding markets using technology first. And I can see the merits of that. It’s not us because we know a part of what we provide that is valuable and differentiated is human expertise.
So that is always going to be part of what we do. We began, so we’re in 35 cities now, we’ll be in
43 by the end of the year. We began in mostly major metropolitan areas. So you would have two or three reporters. But again, our goal is to be able to be in hundreds and thousands of communities. And most of those do not look like Chicago and Denver. They are smaller. They have different makeups. And so we knew we needed to find variations of the model that would allow us to be in smaller communities.
So we’ve now launched some one reporter cities. Our first one was in Boulder and it’s doing incredibly well. And we are exactly and that’s deliberate. That was thinking regionally. So we’re trying to figure out again the mix of human and technology. So the advantage of being in a region is there are some small scale benefits of scale where you can back someone up because of that.
Pete Pachal (10:22.709)
That’s kind of almost gender.
Allison Murphy (10:42.05)
Boulder reporter is on vacation. You still have to put a newsletter out. So there are human solutions for that and there are technology solutions. This year, we’re also beginning our first, I’ll call them half reporter places. And again, one of these is in Denver. So in Arapahoe and Douglas counties, we’ve hired one reporter who will cover both of those counties. Those counties will have their own newsletters. It’s not a combined newsletter. And we’re beginning that as twice a week they’ll get that newsletter. So it’s not every day. Our expectation is that we’re able to bring it to every day using that one reporter by being able to utilize other content from the region and by supplementing with technology supported content and planning and efficiency tools that allow that one reporter to do more.
Pete Pachal (11:38.478)
So how do you, this might even be zoomed out a bit more, but like, how do you decide even to go into a community? What’s the criteria? It seems like it’s probably a mix of, well, like you mentioned, you’re not looking to replace like the local, like the major local papers and that kind of thing. But there are obviously opportunities outside of those. is it kind of like, can find like the criteria of like, okay, there are stories to be told here and people aren’t telling them.
You know, but whether, you know, maybe Boulder’s a good example of that because it is, you know, such a hotbed of startups and all this kind of other things, or at least it was, I haven’t checked lately, but, you know, like, and then, you know, can you reasonably expand? there people close by to back up? there economics? You know, obviously it’s all of these factors, but like, maybe give me the top three and like how you sort of filter that.
Allison Murphy (12:18.721)
Yeah.
Allison Murphy (12:30.092)
Mm-hmm.
Sure. I wish I could say it’s more scientific than it is, but the interesting challenge with running a local business is there is so much opportunity that there almost aren’t bad choices to make. Our cost structure is at this point low enough that pretty much any area of, I don’t know, 50 or 100,000 people has the economy that could support what we need, has enough advertisers and people there. It’s also not an issue of competition because unfortunately, there just isn’t much anymore, right?
There are so few options that even in a place where there’s still a thriving local newspaper, where there may be other smaller local digital properties, there is still so much unmet need that we know we can be there. So we initially, in our first couple of years, went to 30 places. And those were primarily the top MSAs in the country, except New York and LA, because they’re enormous and they are definitely not underserved. So we knew that those were outliers. And we thought, if we’re trying to prove something different, let’s go right below those.
Now we are looking regionally and that’s where our focus is over these couple of years to see how that changes the ability to serve more communities in that area. Because to your point, Boulder is sort of Denver, but don’t tell that to people in Boulder, right? Or we’re launching in Scottsdale and Scottsdale is not Phoenix and don’t you forget it. And like that matters. It matters to us exactly how our readers see themselves. that said, there is absolutely news and information that is relevant, whether it’s related to things going on at the state. There’s so much business-wise that spans a region. And even if you look at something like what’s going on with schools, what happens in a large neighboring area?
This absolutely influences what people are talking about and thinking about. We now are looking very much regionally and we’re building out regions right now around Denver, around Phoenix, around Miami, looking at how we have different combinations of communities and reporter resources to figure out the ways we can flex it.
Pete Pachal (15:03.735)
So let’s talk a little bit about the actual operations, the newsroom operations. OK, you decide to spin up one of these communities, get a newsletter out there. What does it look like? Again, obviously focusing on, well, how do you use AI to make this as efficiently as possible? We can get into the axiomizer and how that factors in. What does it look like?
Allison Murphy (15:25.613)
Yeah.
Allison Murphy (15:30.711)
Yeah. The fascinating and troubling thing about this is that the hardest part is still hiring the reporter. The human element is the first thing we have to do is find the right reporter for the area. And you would think like, okay, well, my gosh, given the decline of newspapers, there must be so many incredible journalists out there. And that is true. But this comes back to the fact that like what we’re asking is hard. Like you’re going to be the only person putting out your newsletter.
Pete Pachal (16:06.349)
Of course.
Allison Murphy (16:07.918)
And we have a very specific style at Axios called Smart Brevity. So like that is our reporting.
So we do ask a lot and need a lot because if you’re the only reporter we also have to nail it. You have to know that community. You have to understand its vibes. You have to be well sourced. You also can’t just cover one beat. You know like maybe you were a fantastic City Hall reporter but now you have to be able to look and not report on everything but understand everything to then be able to say these are the things that matter. So the first and most important part is getting the hire right.
Technologically now, we can spin up a newsletter basically with the push of a button. Newsletters have been part of Axios for a while, so that part’s easy. What we’re doing now is building different marketing technology that allows us to build out and scale audience quickly. So we are, and hopefully we’ll have more to share about this, but we’re beginning to think about how we can use agents to build many marketing teams because
We can’t just keep hiring marketers for every new city we want to launch. And again, our aspirations are hundreds and thousands, not dozens.
Allison Murphy (17:20.396)
We also are looking at ways we can use AI for training reporters. This will be announced? I’m not sure when this podcast goes live..
Pete Pachal (17:34.423)
Couple weeks.
Allison Murphy (17:48.481)
okay, a couple of weeks. So we have an exciting partnership that should be announced by then that I’ll follow up with you on that is working with a technology company and a nonprofit organization to build a scaled training approach that trains journalists and will be free on the use of AI as well as journalistic skills. So that’s part of trying to solve that pipeline problem.
Pete Pachal (17:58.638)
Hmm.
Pete Pachal (18:02.413)
Well, I’m currently right now confronting that I’m being replaced by AI. That’s what I do. I do some journalism training, but you know, it’s okay that you went to AI.
Allison Murphy (18:07.438)
Well, yeah, it’s, yes, and we’ve talked about that. Yeah, you’ve built a, we’re not quite at the comprehensive level you’re creating, but we’re very targeted. And within Axios, we’re looking at how.
Pete Pachal (18:20.481)
yeah, I’m sure it’s very targeted in terms of what we do.
Allison Murphy (18:26.922)
AI helps us create better training and more personalized training too. But also when we’re thinking new cities, part of that is reaching audience as quickly as possible in a lot of places. So we have just been turning on a lot of automated social distribution that changes so that our reporters who used to have to spend like 45 minutes to make an Instagram post, they would have to like go into Figma and resize an image and like come on, we just can’t be doing that in 2026. So we have built, we’ve taken some third party tools and layered in our own additional context and prompting to amp them up so that now it’s like a two minute process to have any story go on social. And that means every city will now be on Instagram, X, Facebook, by the end of the year with every story we publish.
Pete Pachal (18:58.605)
Of course.
Allison Murphy (19:24.318)
So there’s a lot of different ways that the technology is helping us go to a new place.
Pete Pachal (19:28.065)
That’s great. I’d love to zero in a little bit on the AI use plus hiring. So I thought about this as you were talking about that. Certainly thought back to the sort of mini flash in the pan controversy of the Cleveland plane dealer. And I’m sure you read about this and, you know, I hate to keep harping on it, but it was sort of like this moment of like, they were hiring and they made clear to the person they were hiring that they use AI in their process and they described.
Allison Murphy (19:32.877)
Yes.
Allison Murphy (19:44.686)
this.
Mm-hmm.
Pete Pachal (19:58.286)
how they had sort of a rewrite bot, et cetera. And this became like, you know, kind of a bit of contention with that particular candidate, which inspired the column that they wrote about, et cetera. That’s the quickest recap I guess I could give. the, so I’m curious about that. Like in terms of like when you’re looking for people and the AI, because it’s gotta be a different job today than you would have been as a local reporter even, you know, a year ago, right? So is…
Allison Murphy (20:01.443)
Yep.
Allison Murphy (20:07.649)
Mm-hmm.
Allison Murphy (20:22.327)
Yes.
Pete Pachal (20:25.639)
Again, I know you’re not the hiring manager of everyone, it’s like, in terms of, it easier, harder to sort of explain that process? And maybe we should get into a little bit more of the process as you explained that, because it’s like, you know, the thing I just sort of cited was a very specific use case of, something writing articles around the reporting with reporter approval. I’ll, I’ll add. so, so explain to me like how that…
Allison Murphy (20:39.628)
Mm-hmm.
Allison Murphy (20:50.178)
Yes, yes.
Pete Pachal (20:54.689)
…might factor into what you’re doing? That part of what you’re doing? Is that what the axiomizer does? I don’t even know. And then how does that sort of loop back into the type of person you hire?
Allison Murphy (21:05.44)
Yeah, yeah, lots to unpack there. And again, my dog seems to be giving us a symphony in the background. So apologies to all the listeners. Again, so much construction here. There’s a few things. I think one thing, and this isn’t to point fingers at any other publication.
Pete Pachal (21:10.381)
He’s very excited about AI.
Allison Murphy (21:31.477)
I think we at Axios have been incredibly vocal and transparent and consistent in how we think about AI for three years now. I mean, since ChatGPT is coming out. And I do think there’s a difference between talking about things in the JD and really making it clear in every conversation the degree to which we believe this technology is and will be central to every job at Axios. And again, that’s not to point fingers at any publication. We’re all trying to figure out a wildly uncertain world and make our ways along it.
But I do think we are, so upfront about that that it helps a bit in self-selection about who wants to be here and that’s fine. We want that. The other piece is we have said from the beginning that for local…
Technology has to be a part of the solution because the core issue with the sustainability of local journalism is the finances. The math doesn’t work. You have to find ways to keep costs low with quality high. And that cannot be achieved at scale just with humans. One of the things that we’ve seen and that has also been part of are newsroom and our reporters not just being on board but being excited about what we’re doing with technology is it is making their lives better. So we go back to when we launched Axios Local five years ago and you are writing that newsletter every day you are opening up and you see a blank page and you’ve got to fill it by the next day and it has to go out and that happens whether your dog’s barking or your kids are sick or you have vacation next week like it has to happen.
Allison Murphy (23:31.511)
Now we are building a system that is going to create a feed where you can say, here are five recommended items that could go in your newsletter based on what you’ve put in your reporter planning tool, based on data sets that are showing metrics that spike in your community, based on events that readers have sent into you. And you’re able to look and say, and say, all right, wow, here’s four that I’m starting with.
And now I need to think about what the one big thing for the day is. So that is meaningfully changing how reporters are able to do their jobs in a sustainable way and put their time and effort towards the most important reporting that only a human can do. So I think that that’s exciting.
Is it different than what… a local newsroom 20 years ago did. Absolutely, but like candidly that’s not an option, you know? So continuing to work that way, if that’s not on the table, what we try to do and are doing is to paint a picture that’s like, here’s how you can continue to do great original on the ground reporting…
Pete Pachal (24:30.317)
Hmm.
Allison Murphy (24:44.79)
…and bring tremendous value to a community. And these tools are going to help you do that in a way that allows you to take a vacation while still doing a job you love. And again, we’ve seen primarily excitement. have had multiple, I mean, we have ongoing AI projects that are led out of the newsroom.
Pete Pachal (25:10.507)
Nice.
Allison Murphy (25:10.69)
We have a local AI champions group who are building prompts, building tools. We just had a reporter build entirely on her own a data visualization customization tool that will now spit out charts for you give it a data set, like say home price changes. And that’s something that is data we can get from real estate companies. So you have the data set, it’s showing the top 500 areas in the country. It’s in a spreadsheet and it will create the bar chart for every single city. And then paired with AI prompting, it will even tell you if you are in, you know, Austin versus Houston, what’s the data headline? What’s the difference between those areas?
Pete Pachal (26:00.65)
nice.
Allison Murphy (26:03.373)
And that data chart customizer was built in codex by a reporter over the course of
a week.
Pete Pachal (26:04.225)
Custom, it’s not the customizer.
Allison Murphy (26:06.958)
Right, we need a better name for it. Right now we’re just saying it, you know, makes… We do have a localizer, so maybe this will be the local visualizer. We gotta work on the naming.
Pete Pachal (26:15.446)
Right, right.
Yeah, you know, and I guess in Britain it would always be with S’s instead of Z’s. But, well, that’s, that’s cool. Like I, I’m glad you mentioned Codex, because I know Axios, you guys have a licensing deal with OpenAI and I believe as a part of that, usually most of these deals have sort of a trend, like you also get access to chat, GPT and other OpenAI tools, right? And I’m just curious on terms of that deal, like it sounds like it’s, convenient to, but not necessarily required. Like you probably have other tools that you’re using in your
Allison Murphy (26:22.164)
Exactly. But we like our Z’s. Our Zeds.
Yeah.
Allison Murphy (26:36.258)
Yes.
Pete Pachal (26:47.541)
in your whole AI tool set, I assume.
Allison Murphy (26:49.622)
Yes, we use everything. We do have a fantastic partnership with OpenAI and that partnership has been critical to where we are now. First of all, they supported us going to more cities, which meant that we had new pure test beds for some of the things we wanted to try. But we also, a year and a half ago, were able to give enterprise access to every employee, which means we are farther ahead than certainly any media company I know. I mean, you told me, you’re expert on this, but we had every employee go through 101 and 102 and security training well over a year ago now. And we have 90 % of our employee base using ChatGPT, you know, on a weekly basis. so we also have amazingly a lot of tokens as part of that deal.
Pete Pachal (27:35.863)
Nice.
Allison Murphy (27:47.745)
And so in a lot of ways, when we’re trying things out, we default to OpenAI tools because we know them as a company best, but we’re trying everything and we absolutely were trying Claude in different ways. And then once you get into, again, Codex versus like cowork or Claude code, we are going through, again, this is all happening in real time. We talked about this yesterday at leadership team. Who should have access to those things? Because the security risks become different. We need a different level of education.
So though everyone at the company could be opening up Codex, do we want everyone doing that? And our answer is sort of, but once they get a little bit more training. So we have worked internally to identify, yeah, we’ll cut that now. You may be getting a call because we have those needs and you know better than anyone that like the industry can’t keep up.
Pete Pachal (28:33.997)
I get my email is public
Allison Murphy (28:48.99)
Like we’ve reached out to our tech partners and said, what do you have that helps us understand the specific security concerns for a newsroom and how we should think about that. And there isn’t anything yet. There will be, but there isn’t now. And so we’re creating certain things and looking for good partners.
Pete Pachal (28:49.292)
Right.
Pete Pachal (29:08.171)
Yeah, those agentic tools, you mentioned security and scape guard. Certainly it’s less about like what, who has access to it and more what the tool has access to and at what level. And that becomes a very complex thing when you’re dealing with both direct connections to apps through APIs and MCP servers, as well as computer control, right? Which could stimulate a human. So it gets messy. Obviously you can disable those and…
Allison Murphy (29:18.56)
Exactly.
Allison Murphy (29:32.459)
Exactly.
Pete Pachal (29:37.813)
…do go in baby steps, which I always recommend, but the, that, know, there’s, there’s a certain amount of just, you know, increased convenience and efficiency versus increased risk. And it’s always about minimizing that and balancing and training. So.
Allison Murphy (29:49.739)
Absolutely. we’ve, yeah, and we’ve, we’ve found and you know, we’ve got 400 or so people at Axios. So that’s not a massive company. But it is big enough that
This isn’t easy. It’s not easy anywhere. And we found that segmenting employees, partly by self-selection, has really been the key. So even when we got OpenAI, we said we’re looking to create a set of champions with people from each department who are excited about this.
Pete Pachal (30:25.461)
And now some of those people.
Allison Murphy (30:26.954)
Yeah, which was more than maybe what we wanted. So we ended up picking, you know, two or three per every.
We had a bit of a problem because we had 100 people raise their hand to do that, so 25 % of the company, which was a team of a certain size. Now we have also just organically seen builders and super builders emerge. Some are people you would expect like people who are deep in the process and operations of like our sales team where they see ways we can get more efficient. But we’ve just moved one of our employees who was a fantastic individual contributor on our finance team. She was doing such incredible work automating aspects of our financial processes and got so jazzed about it that we were like, you know, we kind of need another person internally who is focused solely on employee enablement of AI. And she’s just taken that on full time. So she has left her finance job and is now our like internal AI trainer and enablement lead and is working with other employees to help them say, how can you in a self-serve way solve some of the workflow problems you see?
And that’s another way that we’re trying to kind of unleash a next layer of employees is like with assistance with someone who’s trained but also with someone who you know isn’t like did this herself like she she has gone down this path and has has figured out what it means for someone who doesn’t have a technical background to go through a process of how you can use these tools to make like your life and our business better.
Pete Pachal (31:58.52)
Sure, sure. So listen, just to clarify on the bringing it back to the newsletters, particularly the local newsletters and the Axiomizer. So as you described it, there’s a scanning sort of media monitoring, beat monitoring, region monitoring aspect to it. You know, clearly figuring out what’s going on in the community, summarizing and writing. Like I want to just be clear, is the tool writing copy that the reporter then approves for?
Allison Murphy (32:23.052)
Yeah.
Pete Pachal (32:28.641)
publishing or is it just research that helps them right?
Allison Murphy (32:31.01)
No, that’s not what the Axiomizer is doing now. Now, we are building deeper technology that will create partial drafts, but not complete, and that’s not in use yet. And that will never happen without it going through a reporter. So it is not our intention to be creating content that just goes out having never seen the light of day.
Pete Pachal (32:42.828)
Okay.
Pete Pachal (32:57.899)
Of course. Yeah. And I didn’t mean to imply that it was not approved or edited. It’s more than that. Is it copy? Yeah.
Allison Murphy (33:00.27)
No, no, no, no, it’s your right to specify. Your right to specify. And this is what we are very deliberately trying to understand where our boundaries are, both what they technically can be and what we think they should be in terms of the brand promise and the trust we have with readers.
So what our tools can do now is take drafts and copy that reporters have done and help really go through and identify ways you can make it more in Axios style, flag areas that maybe need fact checking, give recommendations on areas that can be stronger. So think about it as a really kind of souped up editor. What we are also building, as I said, are ways that the some of the drafting at least can be done in a more automated way. So where we’re beginning with that is with what I called the localizer. Again, we need the names, but some of what our reporters do now is look to see what national stories could be relevant for their city or what other cities are writing that could be relevant. Maybe that’s just for a story idea or maybe it’s something where like, you know, 80 % of it is pretty relevant as written.
Pete Pachal (34:01.645)
Mm-hmm.
Allison Murphy (34:20.256)
What the localizer can do is go in and say, OK, here’s the story. And it can both flag and highlight and recommend copy that can make that story relevant for, again, let’s say, Boulder. So I used the real estate example. You’re still going to have to go in. If you put that story out as is to a city, it would be really boring.
It would be like a statistic about home prices. So reporters are still the ones saying like, OK, does this even matter for my readers? If so, what makes this complete? I need to go do some sourcing. I need to talk to people in the industry. I need to talk to homeowners. I need to understand what turns this from a data set that says Austin instead of Houston into a piece of journalism that’s worth reading.
But rather than that reporter having to go find the data set on home prices in Austin, build the chart, create the comparison to other Texas communities, that work can be done for them. And backing up, and this is kind of the next set of the technology, we can also push that to them.
Allison Murphy (35:36.277)
If it’s data we’re managing centrally to say, Austin is spiking in this data. Because you know what, if Austin’s at the median, that’s not that interesting. If Austin’s in the top 10%, that’s more interesting. So how can we say to a reporter, all right, we’ve got 50 different data feeds. in the updates from the last month, these 10 things, it looks like there’s something going on in Austin. Those might be interesting stories for you to look at.
So yes, we can create automated charts and we can do some of the initial drafting around the data. But like a machine’s not going to tell you whether a reader in Austin cares about something, a reporter who is of that community and understands it or who can go and speak to members of the community because they are deeply sourced can do those things.
Pete Pachal (36:15.831)
Mm-hmm.
Pete Pachal (36:24.669)
Do readers have any, well, what’s your experience with like, how do readers feel when it does come to AI? And I suppose that has to do a little bit with your disclosures and where you would even, because in terms of research and putting things together, I don’t even know if you need to, right? And you tell me if that’s your policy. But are they mostly indifferent? Are they just kind of like, well, it’s just good stories and I like the stories? Or did he feel like they sort of react?
Allison Murphy (36:39.927)
Right.
Allison Murphy (36:49.197)
Yeah.
Pete Pachal (36:53.599)
I mean, again, I’m thinking in my mind about newsletters, but obviously AI also opens the possibility for other formats. So tell me how you’re thinking about that in terms of reader reaction and reader experiences.
Allison Murphy (36:59.2)
Yeah.
Great, great question. At this point, we still have so much deep reporter involvement that, again, we’re using AI in the kind of planning and preparation. But everything still is quite human in terms of reader response to us specifically, we haven’t seen a whole lot of that yet.
Pete Pachal (37:12.791)
Right.
Allison Murphy (37:28.448)
We do know because we our audience tends to be professionals for local, but also especially for our national audience. We write about AI a lot. And among business and policy leaders, there is endless appetite to know what’s going on. Now, what does that mean when you start to talk about AI in the content? I think we don’t totally know yet. And this is a place where we intend to do some really deliberate and very transparent experiments. And I will definitely let you know, Pete, when some of those are live, because I think we’ll have some in the next quarter, where at Axios, we believe, bar none, that you have to have expert reporters. If you don’t have humans who are expert in what they’re doing, you can’t create something that’s differentiated and that therefore has a business behind it. So like an AI-only news organization.
Pete Pachal (38:01.026)
Yes.
Allison Murphy (38:23.15)
Like good luck the LLMs are just gonna do that, you know, like everybody will just be able to ask Claude for their update soon. So like we can’t that’s not a future for us we also believe tech can help us serve more readers and Help us bring news and information that otherwise maybe we couldn’t to those readers Where that balance is that can like best serve our readers?
Pete Pachal (38:26.849)
Mm-hmm.
Pete Pachal (38:31.33)
Ha ha ha.
Pete Pachal (38:35.116)
See you.
Allison Murphy (38:52.236)
We don’t know yet. So the goal of some of the experiments we’ll be doing will be to be in like very open dialogue with readers. So imagine a universe where we’re saying, okay, we’re going to do something different with AI. We’re making this totally opt-in for you reader. You’re basically going to be the voluntary beta tester for some experiments with this. And we’re going to really reach out and hear from you about what you like and what you don’t. And it’s going to be incredibly transparent what we’re doing because we’re going to tell you. I see this.
And they can tell us, right? Because at the end of the day, if the readers don’t trust us and don’t like what the product is, then we don’t.
Pete Pachal (39:24.287)
Mm-hmm. Yeah. Well, and they can tell you, I guess, just by hitting Reply, right?
Allison Murphy (39:34.944)
We don’t have anything. It doesn’t matter how much we believe in AI. It matters how it’s serving our readers. But like you, we’ve seen… sorry. Go ahead.
Pete Pachal (39:40.439)
So that brings me… I just wanted to… Actually, I’m interested in what you’re about to say. Like me, you’ve seen what?
Allison Murphy (39:47.277)
Yeah. well, you’ve seen, we’ve all seen research about the tremendous skepticism and fear about AI writ large, which I think is justifiable. But that is also talking about broad population on AI broadly. We have a specific audience and a specific product and I don’t think those things necessarily go hand in hand. And I don’t think it is enough for the industry to say, the industry, meaning news, to say, people just don’t trust AI and to therefore step away from it. I think that is a path to nowhere.
I do think we have to be responsible and transparent and be open to the places where we may get it wrong. And I don’t ever want a reader to be surprised by something that’s in front of them and how it got there. So we’re being very, very rigorous in that. And one of the things that we think is going to help us be smart is, again, some very structured experiments that are opt in and that have a high degree of reader feedback, like baked in, to help us get smarter on what our audience wants and how specifically for Axios, we can or can’t or should or shouldn’t be using certain tooling.
Pete Pachal (41:13.302)
Well, speaking of roads to places, let’s think about a road to somewhere and sort of deliver on what I promised at the outset. Give us the snapshot. How are you going to know this is all working beyond the obvious?
Allison Murphy (41:24.878)
Yeah, a few different things. One, we already know some things are working because when we look at individual tools, like I talked about the social automation, it is incredibly apparent the difference between 45 minutes and two minutes to create a post. So when we look at individual tools, we look to say, wow, is this an order of magnitude better?
The next step and the next thing we’re looking for is I mentioned these one reporter and half reporter cities. Are we able to create a good product with where we are now? And are we able to then increase the number of sends, know, go from a twice a week to three to four to five times a week newsletter by bringing in more technology? So that’s a thing we’re looking at.
Most of these experiments are existing on a project level and we believe we need to do lots of them. Some won’t work, but we think enough will that then they come together in a way that that can work. So some of those are about reporter efficiency. As I said, some of them are about how the business works.
I mentioned we’re on the very early stages of thinking about how agents can help scale marketing. We are also building some very cool sales tools that do automated prospecting and can create custom ads to send out on spec to potential clients. So on the business side, we’re absolutely building those too. So for every project we’re looking at, what are the gains we’re getting. And then we know for local, for instance, we know the economics of how much we think a city can cost and what the revenue ramp looks like. And as those numbers come into, come into balance, we’re going to keep adding cities. So you’ll know it’s working when you see us announcing more and more places.
Pete Pachal (43:26.252)
Nice.
Pete Pachal (43:31.071)
So I try to close out these conversations in a similar way most times, and I obviously want to keep things forward looking. You’re looking out into an AI mediated ecosystem that’s happening in the next few years. When you look out at that, what are you concerned about? Give me one thing you’re concerned about, and what’s one thing you’re really hopeful about?
Allison Murphy (43:54.831)
I’ll try to keep it focused to, again, news and media because otherwise this becomes a big conversation. What concerns me is the pace of change.
And that’s for a couple reasons. One is I think a lot of news organizations simply don’t have the cash in hand to weather the speed at which changes are happening. And so I think we’re going to see a lot of organizations that might be quite strong or are differentiated and are doing good work that maybe just like can’t get through the survival curve as we move from a programmatic volume traffic-based business into something different. So the speed of that concerns me. also wonder about like audience discovery and growth. How are new people going to find us? And that’s not just a media question. That’s a question for every brand.
And again, goes back to kind of a cash question of like if the only way to acquire audiences will be by paying for them on other platforms, like that takes money in an already margin thin business. So those things are concerning to me from a business model standpoint from the industry. What excites me is I think the ability of readers to get more news and information will probably increase, you know, like there’s just more more ways. Like right now I think we’re all inundated with information, but a lot of it is just not that good. So I think that this is is AI gives more tools for people to get the kind of information they need and want and that also will help.
Allison Murphy (45:48.682)
news organizations do the same internally. It’s going to help us figure out more about what our readers need and want and bring it to them. And for local, like we are already seeing this, the ability to bring costs way, way down. I mean, we are absolutely on the cusp of a financial place where the only cost of a new city is a reporter and then some small portion of like sales and marketing capability. And that is incredibly exciting if you think about being able to have skilled trained journalists in every community in America.
Pete Pachal (46:24.769)
Nice. Good stuff. We’ll leave it there. Alison Murphy, thank you so much for dropping by and sharing with us what’s going on at Axios and Axios Local.
Allison Murphy (46:28.525)
Yeah.
Allison Murphy (46:33.218)
Thanks for having me.
